TV Ad Frequency-Response Profiling for Optimal Ranges
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The challenge in TV advertising is measuring the impact of linear TV commercials on website traffic due to the natural distinction between online and offline mediums, creating a data gap that makes it difficult to correlate and accurately attribute website visits or app downloads to specific TV spots.
Innovation Solution
A system that processes IP-level impression data using a household device graph to determine frequency-response profiles, correlating impressions with responses and conversions, and applying an optimal frequency algorithm to identify the best frequencies for maximizing response rates, thereby guiding advertising campaigns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional Nielsen audience measurement systems are used, then audience size and composition can be measured, but the ability to attribute website traffic to TV commercials is lost due to the data gap between online and offline mediums
Solution Approach 1:
The patent introduces an intermediary system that bridges the data gap between offline TV advertising and online website traffic. This intermediary uses IP-level impression data, household device graphs, and attribution models to connect TV spot exposures with subsequent website visits, enabling measurement of the previously unmeasurable attribution relationship between linear TV and digital engagement
2Measurement precision
If frequency-response profiles are generated using comprehensive IP-level data processing, then optimal frequency ranges can be accurately determined, but processing time and computational complexity increase
Solution Approach 1:
The patent segments the complex frequency optimization process into distinct manageable components: (1) collecting IP-level impression data, (2) building household device graphs, (3) calculating frequency-response profiles, and (4) determining optimal frequency ranges. This segmentation allows each component to be optimized independently and enables parallel processing, reducing overall computation time while maintaining accuracy
Solution Approach 2:
The patent performs preliminary actions by pre-processing IP-level impression data and pre-building household device graphs before the main frequency optimization calculation. This preliminary data preparation and structuring reduces the computational burden during the actual frequency-response profile generation, significantly cutting processing time while preserving measurement precision
Data Source
AI summary
A method for determining an optimal frequency range for maximizing the response rate may include Internet Protocol (IP)-level impressions from impression data, rolled up the IP-level impressions to household-level impressions, and then correlate the household-level impressions with response and conversion rates so as to determine impressions with associated attribution. Each impression is assigned to a frequency bin based on a rolling frequency calculation. Relative net funnel efficiencies (NFEs) can then be determined by comparing different frequency bins, each respective frequency bin containing frequency-specific impressions, a frequency-specific response rate, a frequency-specific conversion rate, an NFE determined from the frequency-specific response rate and the frequency-specific conversion rate, and a relative NFE for the respective frequency bin. A frequency-response profile containing the relative NFEs is generated and provided as input to an optimal frequency algorithm so as to produce an optimal frequency range which, in turn, is displayed on a client device.